Dick Olsher Dahlia speaker Dick Olsher updates

The Dahlia Debra: Dick Olsher updates his DIY loudspeaker design, from June 1987 (Vol.10 No.4):

Since the publication of the Dahlia do-it-yourself speaker article in February 1986 (Vol.9 No.1), I've been inundated with requests for clarification of construction details and additional information. I'm very pleased that the Dahlia has been received so favorably and wish I could personally acknowledge each and every one of your letters. To have done so, however, would have kept me very busy for a long time, and would have precluded additional research work on my part. Instead, let me take this opportunity to respond to the questions raised most frequently and to describe an exciting upgrade to the Dahlia. First, there are no construction plans available from Stereophile: what you see is what you get. Yes, JGH implied there might be such plans, but he was actually referring to the construction details provided in the original article.

The Enclosure
Two minor clarifications are in order. The crossmember of the "H" brace does indeed run from front to back—as shown in the top view. The distance from the top of the cabinet to the center line of the tweeter was inadvertently omitted, and is 4.75". My original plans for the Dahlia called for a 20&$176; tilt-back of the front baffle, not only for time-alignment purposes, but also to optimize the tonal balance in the upper mids. Without the sloping front, the radiation pattern exhibits a slight suckout around 3kHz. The tilt-back provides better symmetry in the radiation pattern and results in a neutral midrange balance.

Rather than complicate the woodworking, I opted for a simple box with a tilt-back stand, only to realize after the fact that appropriate stands are hard to come by. As some of you discovered, the Chicago Speaker Stand (model RJ100) that I recommended is no longer available. I have resorted to using a particle board base under the Dahlia with a 15° tilt-back. This base is then positioned on top of a 15" tall Chicago Speaker "Hercules" stand, as shown in the accompanying photograph of the Dahlias. A 20° base would be even better, but then the entire assembly becomes unstable and the Dahlia ungracefully flips over backward.

In retrospect, the best approach would have been to bite the bullet and embrace the sloped cabinet, and this is my current recommendation for those of you with the woodworking savvy. Just be careful to maintain the present internal volume (1.36 cubic feet) and front-baffle dimensions by increasing the depth.

There were several questions concerning alternative cabinet materials. Peter Crosby of Claremont, New Hampshire, has the best approach. He is apparently planning to cast the cabinet out of concrete, which is eminently rigid and damped, and is expecting nothing short of "concrete" results. The initial investment in researching and perfecting the proper molds is, however, quite considerable—at least compared with the cost of the materials involved. Peter is expecting the final cabinet to weigh from 100 to 200 lb—not exactly UPS shippable! Sticking with wood, an interesting composite is a sandwich of 0.5" plywood and 0.5" particle board, glued and screwed together. Although more difficult to work with, it is far less colored than a comparable thickness of either material alone. The mass-market approach would be to use 1" particle board or plywood, and this is what I recommend at the minimum.

Diffraction Control
Several readers have complained of not being able to locate any ¼" felt for diffraction control around the tweeter. I have routinely used weatherstrip foam tape for this purpose. The stuff is ¼" thick by ½" wide with an adhesive backing, and is readily available in local hardware stores. At least three rows of tape should be glued around the tweeter faceplate. It's not a bad idea to leave about a ¼" between rows, which increases the HF-absorption efficiency of the overall pattern.

Wool Stuffing
Be careful not to overstuff the cabinet—this is not a Thanksgiving turkey! Contrary to what the folks at the Rio Grande Wool Mill may tell you, the proper amount of long-fiber wool is as I originally indicated: one-half to one pound maximum per cabinet. I have experimented with as much as one pound, but feedback from the field seems to favor the lower limit. The danger in overdamping the enclosure is of choking off the vent contribution. The bass-reflex action diminishes as the cabinet becomes more lossy.

The Crossover
I may have been a bit understated in extolling the virtues of the SiderealKaps. These are still, by a large margin, the best caps for the Dahlia crossover. The improvement in treble clarity and detail is very noticeable compared with the French FRC (Solen) metallized polypropylene caps. Unfortunately, the Sidereals are very expensive; many budget-minded audiophiles may find the cost prohibitive. Also irksome is the Sidereal's tendency to be on the low side of the nominal capacitance value. For example, a 3µF cap may measure 2.8, still within the ±10% spec, but for this expensive a cap the tolerance should be no worse than 5%. Be sure to measure the caps you buy, and beef up the values to within 5% for the best results.

The best coils I've found so far are Madisound sidewinder coils. These are 16-gauge air-core type with very low DC resistance values and very good tolerances.

The Woofer
Audax claims that they've experienced serious production difficulties with the TPX drivers and have been unable to meet initially advertized parameters. For example, the original TX-2025-RSN was touted as having an overall Q of 0.46 and a free-air resonance of 38Hz, while current production is typically of the order of 0.6 for Q and 49Hz for the free-air resonance. The effect of all this on the bass alignment of the Dahlia is to increase the response peak at 60Hz to 4dB and decrease the half-power frequency to 55Hz.

This prompted me to take a closer look at the Audax catalog. The TSN version of this driver, which sports a slightly larger magnet and thus a lower Q of 0.5, appears to be a better choice—it gets much closer to the original bass alignment. The bass peak with the TSN version should be no greater than the 2.5dB originally intended. I asked Audax to send me samples of the TSN cones for evaluation. What I got, however, were more RSNs, so as yet I have no final report on the matter. Since the TSN and RSN versions are identical except for the Q parameter, I recommend trying the TSN for new construction. I would not personally go to the trouble of modifying existing Dahlias, as the sonic differences should be small. In either case, be sure to use the 8-ohm version of these drivers.

The Dahlia Debra
I've left the best for last. You should recall from my original article that I concentrated my design efforts on midrange excellence. Subsequently, I was forced to take a closer look at the upper octaves. It's not so much that the Audax dome is bad—it is, in fact, a creditable performer, but it is a tad too bright, sluggish, and fuzzy compared with the best domes money can buy. I began a systematic search for an alternative tweeter for the Dahlia.

Candidates would have to meet three important criteria: sonic excellence, physical compatibility with the existing front baffle cutout, and no crossover changes. The latter two criteria, of course, were aimed at facilitating the upgrade of existing Dahlias. I looked at tweeters from Focal and Dynaudio, as well as other expensive domes.

The winner is: the MB Electronics model MCD-25M titanium dome. This is a West-German 1" dome of very good construction quality and excellent specifications. I sat on samples of this dome for almost a year because of negative feedback I had received in the past about metal domes. Such domes are supposed to be out of control in the time domain, with severe ringing in the ultrasonic range that translates to harshness and sizzle in the treble. When I finally got around to testing it, the MB titanium dome proved to be very well behaved, with no nasty peaks out to 40kHz. Treble response is flat to 25kHz, and rolls off gently to 40kHz. The MB dome is a perfect fit in the present Audax front baffle cutout, and works very well with the old crossover, providing a simple impedance-equalizing network is added to the high-pass filter section. This network consists of a 0.56µF cap in series with a 7-ohm (5 watt) resistor, and is soldered directly across the tweeter input lugs (see fig.3).


Fig.3 Dick Olsher Dahlia Debra, high-pass filter.

I have dubbed the end result the Dahlia Debra, borrowing my daughter Dahlia's middle name to denote the upgrade. This seems more elegant than such mundane designations as "Mark II". The new tweeter dramatically improves the treble quality, with faster and cleaner transients. There's more focus and detail above 3kHz than before. By comparison, the old Dahlia is smeared and veiled-sounding through the upper octaves. Sibilants, for example, are reproduced oh-so-naturally, without gratuitous spit or splash. Despite the clear superiority of the titanium dome, however, it may be an unwelcome guest in some situations. It is much less forgiving than a soft dome, and therefore more revealing of solid-state treble nasties or problems in the front end. Be forewarned.

The MB tweeters are not cheap—around $50/pair. As I see it, the Dahlia is now complete. Some readers have inquired about a subwoofer, and the answer is no. I would much rather conceive a three-way design with deep bass response, and I can tell you that such a design is in the works.

One final note. You may have seen advertisements for Dahlia kits from several sources. You should know that neither I nor Stereophile has in any way licensed or endorsed commercial production of the Dahlia, so that I cannot vouch for any of the commercial products. There is nothing wrong, however, with using the design for commercial gain. I think it's a good design and it is in the public domain. As always, your comments are welcome (care of Stereophile).—Dick Olsher